Looking back on the expansion: This time, it's different

The 2007Q1 national accounts numbers were released last week, and it looks like we have indeed dodged that slowdown we were worrying about last fall. And a good thing, too: compared to the 1982-90 and 1991-01 expansions, this current run has some very nice features.

This is one of those long, data-oriented posts with lots of graphs, and it all starts below the fold.

I'm going to use the graphical device I used here (stolen from Menzie Chinn): all variables are logged, and re-scaled to zero at the trough of the  recession (1982Q4, 1991Q4 and 2001Q3). They are then multiplied by 100, so that they can be interpreted as the per cent deviation from their values at the trough.

First up is GDP:

There's nothing particularly special about the rate of GDP growth during the current expansion: it's pretty much following the 1991-01 scenario, and cumulative GDP growth is 10% lower than what it was at this point in the  1980's. But since the 1982 recession was much more severe than the very mild slowdown in 2001, much of that gain consisted of recovering previous losses.

Now let's look at the components of GDP, starting with consumption expenditures:

This looks a lot like the GDP graph, but it's instructive to compare the scales of the vertical axes of the last two graphs. In the first 5 years of the 1982-90 expansion, consumption grew at pretty much the same rate as GDP. In 1991-01, it was 5% slower. But in the current expansion, consumption is running 5% faster than GDP. For the first time in a generation, consumption is driving economic growth. 

Next is fixed business investment:

Again, this expansion looks like that of 1982-90, except that it wasn't preceded by a 20% drop before the trough. And the less said about the 1990's, the better.

The government spending graph shows just how severe the budget problems were in the 1990's:

Now that deficits and the debt are under control, government spending patterns have returned to their previous form.

On to the trade balance, starting with imports:

The 30% jump at the beginning of the 1982-90 episode was a recovery from the 30% drop during the recession; otherwise, there's not much to distinguish the three paths.

Exports is quite another story:

The 1982-90 and 1991-01 expansion both saw strong export growth, but there's been essentially no growth in real exports this time around. This may look odd at first glance: aren't we running an energy-driven trade surplus?

Well, yes, we are, but it's getting smaller. It's important to remember that a consequence of the run-up in exports during the 91-01 expansion was a (nominal) trade surplus that rose above 7% of (nominal) GDP before the slowdown in 2001. Levels like these can't be sustained indefinitely without an appreciation in the exchange rate, and that's indeed what happened. As the CAD appreciated, some sectors (notably manufacturing) were squeezed out of export markets, countering the growth in energy exports; the net effect has been a wash. Although exports are still high (we still have a trade surplus), they aren't driving this expansion the way they did in 1982-90 and 1991-01.

The behaviour of the exchange rate is what distinguishes this cycle from its two predecessors. Here is the real CAD-USD exchange rate:

The story in 1982-90 and in 1991-01 was of an export-driven expansion, fueled by a depreciating exchange rate. This is a familiar scenario; so familiar that some observers appear to have difficulty understanding how an appreciating currency could be consistent with an expansion. But that's a subject for another post; for now, I'll just note that during this expansion, the real exchange rate has appreciated by 30-40%.

The labour market now. Here is what employment has looked like over the last three cycles:

Not much to say there. The Canadian economy has many problems, but job creation isn't one of them: employment rates are at an all-time high. The usual problem has been one of stagnant real wages. Happily, this hasn't been the case during this expansion:

In 1982-90 and 1991-01, real wages went sideways, but they've increased by something like 8% since 2001. Why? An obvious place to start is labour productivity:

Although labour productivity has grown during this expansion, it can't explain the increase in real wages. Output per worker grew even faster in the previous two expansions, but it didn't generate much in the way of increases in buying power for workers. I've  already blogged on this a few times (eg: here and here), so I'll leave it at that.

A quick summary of what is different about this expansion:

1. The exchange rate has been appreciating.
2. Growth is driven by domestic consumption, not exports.
3. Real wages are increasing.

These points are all linked, of course. The increase in the exchange rate prevented exports from growing very much, real wage growth played a role in the increase in domestic consumption, and at least some of the increase in real wages can be attributed to the pass-though of the higher exchange rate on imported consumer goods.

This time it's been different, and in a good way, too. We'll be sorry to see this one end.

Comments

[Marc Lee]

This is fascinating stuff, Stephen.

One thing that seems odd to me is your figure on real wage gains. According to Statscan, average weekly wages in constant dollars grew by -0.8% in 2003, 0.7% in 2004, 1.3% in 2005 and 1.4% in 2006. Early year-over-year figures for 1Q 2007 are all well under 1%.

Andrew Jackson has some more details on this on RPE today:
http://progecon.wordpress.com/2007/06/08/an-update-on-canadas-two-economies/

[Marc Lee]

One other thing is that if labour productivity only grew by 4% since 2001, while wages grew by 8%, that would imply that wages as a share of GDP have been growing. In fact we have seen the exact opposite, as corporate profits have grown to record highs.

[Stephen Gordon]

The factor share point is interesting; I'll have to think about that. For now, I'll just clarify where that graph came from.

The nominal wage series is 'Wages, salaries and supplemental income' (Cansim series V1996473) divided by hours worked (v4391505). These are monthly data, and the series in the graphs are the quarterly averages, divided by the quarterly averages for the CPI.

I also looked at the business sector series for hourly compensation (V1409158) and got a graph that looks pretty much like the one for the broader measure for the current expansion. The only difference is that in the other two expansions, the business sector real wage shows some growth near the end of the window.

[Valuethinker]

Stephen

Remembering David Foot's lectures on demographics from undergrad, is the different factor on real wages labour force growth?

ie in 1980-90 you had the Baby Boom entering the labour force. This, arguably, repressed real wage growth through much of the 80s.

I remember DKF showing a graph in class that showed that youth unemployment in Canada was never higher than it was in 1981.

1991-01 this wouldn't hold up (unless immigration really picked up?). However the 1990 recession was the most severe in postwar history, and so it's at least credible (I would guess) that this meant labour market conditions were unusually slack, for an unusually long time?

Also this was the period of NAFTA, so that could have had a downward impact on real wages, as Canadians (with lower productivity) adjusted to competing with Americans (higher productivity) and Mexicans (much lower wages). Again the problem with that thesis, in and of itself, I suppose, is that the Chinese are a factor this time, at least as great on a (smaller) Canadian manufacturing sector as NAFTA.

Sorry a little of a ramble here, but LF entry would be my obvious first thought.